Automatic dike danger patrol method based on vehicle-mounted unmanned aerial vehicle and user server

Abstract

The embodiment of the invention discloses an automatic embankment danger patrol method based on a vehicle-mounted unmanned aerial vehicle and a user server. The method comprises the steps that a calculation service layer of a self-fusion calculation system obtains an embankment dangerous case patrol task, and the task content of the embankment dangerous case patrol task comprises the steps of controlling at least part of unmanned aerial vehicles to conduct embankment patrol and receiving embankment patrol data collected by at least part of unmanned aerial vehicles in the embankment patrol process; first state information and second state information are obtained, the first state information is used for indicating the use state of the unmanned aerial vehicle, and the second state information is used for indicating available computing network resources provided by a plurality of preset resource nodes; and controlling the target unmanned aerial vehicle and the first target resource node to execute the dike danger patrol task according to the dike danger patrol task, the first state information and the second state information. According to the scheme, collaborative scheduling of the multi-source heterogeneous sensing equipment can be utilized, and global coverage of dike patrol and multi-dimensional collection of data are achieved.

Description

Technical Field

[0001] This invention relates to the field of dike inspection technology, and more specifically to an automated dike inspection method and user server based on vehicle-mounted drones. Background Technology

[0002] Statistics show that my country has nearly 400,000 kilometers of dikes along small and medium-sized rivers. These dikes are extensive, long, and have complex environments. Traditional manual dike patrol and inspection methods are outdated, inefficient, and costly. In recent years, with the increasing demand for intelligent emergency management, geophysical exploration methods such as ground-penetrating radar, transient electromagnetic methods, and high-density electrical resistivity tomography have been gradually applied to promote intelligent dike patrol and inspection.

[0003] While traditional geophysical exploration methods can identify structural hazards such as cavities and cracks within the dike, they suffer from limited effective detection depth, significant blind spots in spatial monitoring, and low operational efficiency, making it difficult to meet the real-time and wide-area requirements of dike patrols. The emerging drone-based dike patrol technology, while offering flexibility and mobility, also faces practical challenges: firstly, traditional drone operation relies on manual intervention (takeoff, landing, and charging), affecting continuous operation capabilities; secondly, the dynamic complexity of the network environment and insufficient communication capabilities in dike areas limit data transmission stability and remote control capabilities. Summary of the Invention

[0004] This invention addresses the aforementioned problems. It provides an automated dike hazard inspection method and user server based on vehicle-mounted drones. This solution enables automated drone inspections using a vehicle-mounted drone nest, and through the collaborative scheduling of multi-source heterogeneous sensing devices, achieves full coverage of dike inspections and multi-dimensional data collection, improving the initiative, accuracy, timeliness, and intelligence of dike hazard response, thus facilitating the automation of dike inspections.

[0005] According to one aspect of the present invention, an automated method for inspecting dike hazards based on vehicle-mounted unmanned aerial vehicles (UAVs) is provided, applied to the fusion adaptation layer of a fusion computing system. The method includes: obtaining a dike hazard inspection task from the computing service layer of the fusion computing system; the task content of the dike hazard inspection includes controlling at least some UAVs to conduct dike inspections and receiving dike inspection data collected by at least some UAVs during the inspection process; obtaining first state information and second state information, the first state information indicating the usage status of the UAVs, and the second state information indicating the available computing network resources provided by each of a plurality of preset resource nodes, wherein at least... Each resource node in a partial resource node configuration includes a network device or a computing network integrated device. Each resource node in a plurality of resource nodes is used at least to transmit dike patrol data. Available computing network resources include available network resources used to support data transmission. Based on the dike hazard patrol task and the first and second state information, the target UAV and the first target resource node are controlled to perform the dike hazard patrol task in order to obtain the dike patrol results based on the dike patrol data. The target UAV is a UAV whose usage status can support the dike hazard patrol task, and the first target resource node is a resource node whose corresponding available computing network resources can support the transmission of dike patrol results.

[0006] Optionally, the dike hazard inspection task includes a drone inspection task and a node transmission task. Based on the dike hazard inspection task and first and second state information, the target drone and the first target resource node are controlled to perform the dike hazard inspection task to obtain dike inspection results based on the dike inspection data. This includes: based on the dike hazard inspection task and the first and second state information, calling a first target strategy and a second target strategy from a preset strategy library. The first target strategy includes a first task allocation strategy, a drone flight path planning strategy, and a drone task execution order strategy for the dike hazard inspection task. The second target strategy... The target strategy includes a transmission strategy for dike patrol results. The first task allocation strategy is used at least to allocate patrol positions for UAVs. Based on the first target strategy, each patrol sub-task of the UAV patrol task is sent to the nest of the target UAV that performs the patrol sub-task to control the target UAV to execute the corresponding patrol sub-task. Based on the second target strategy, the target flow table for node transmission tasks is sent to the first target resource node to control the first target resource node to form a transmission path that can transmit dike patrol results. The dike patrol data transmitted through the transmission path is obtained, wherein the dike patrol results include dike patrol data.

[0007] Optionally, the dike hazard inspection task is represented by computing power identification information and computing power behavior description information. The computing power identification information indicates the dike hazard inspection task, and the computing power behavior description information indicates the computing network resources required for the task. The first state information includes the UAV's flight time information. The second state information includes population identification information, population behavior description information, node identification information, and node behavior description information. The population identification information indicates the corresponding resource node population, and the population behavior description information indicates the preset tasks that the corresponding resource node population can perform. These preset tasks include node transmission tasks. The node identification information indicates the corresponding resource node, and the node behavior description information... This is used to indicate the transmission subtasks that the corresponding resource nodes can execute; based on the dike hazard inspection task and the first and second state information, the first and second target strategies are invoked from the preset strategy library, including: invoking the first target strategy from the strategy library based on the endurance information, computing power identification information, and computing power behavior description information; determining the target group that can execute the dike hazard inspection task based on the computing power identification information, computing power behavior description information, group identification information, and group behavior description information, the target group includes at least two resource nodes; and invoking the second target strategy from the strategy library based on the computing power behavior description information, the node identification information and node behavior description information corresponding to the target group.

[0008] Optionally, the patrol subtask can indicate the target flight path, target attitude, and sensor data collected by the corresponding target UAV.

[0009] Optionally, the fusion computing system further includes a computing network component layer, which includes multiple resource nodes. The computing network component layer is deployed with a first neural network for processing dike patrol data to obtain dike hazard assessment results. The first neural network includes a second neural network. The method further includes: acquiring the mission execution log of the target UAV and adding the dike patrol data and mission execution log to a preset database to update the target training samples in the database. The computing network component layer is used to train the second neural network indicated by the mission execution log based on the dike patrol data in the target training samples; and / or, acquiring the latest available computing network resources of at least one of the multiple resource nodes and updating the acquired latest available computing network resources to a preset database. The database is used to store the available computing network resources of multiple resource nodes. The fusion adaptation layer acquires the available computing network resources of multiple resource nodes from the database.

[0010] Optionally, the method further includes: obtaining a dike hazard assessment task from the computing service layer, wherein the task content of the dike hazard assessment task includes processing dike patrol data to obtain a dike hazard assessment result, wherein each resource node of at least some of the multiple resource nodes includes a computing device or a computing network integrated device, and the available computing network resources also include available computing resources for supporting data processing; and controlling a second target resource node to execute the dike hazard assessment task according to the dike hazard assessment task and the second state information to obtain the corresponding dike hazard assessment result, wherein the second target resource node is a resource node whose corresponding available computing network resources can support the processing of dike patrol data.

[0011] Optionally, based on the dike hazard assessment task and the second state information, the second target resource node is controlled to execute the dike hazard assessment task, including: based on the dike hazard assessment task and the second state information, calling a third target strategy from a preset strategy library, the third target strategy including the target algorithm required for the dike hazard assessment task and the second task allocation strategy, the target algorithm being the algorithm used to process dike patrol data; and sending each assessment sub-task of the dike hazard assessment task to the second target resource node based on the third target strategy, so that the second target resource node can execute the corresponding assessment sub-task respectively.

[0012] Optionally, the target algorithm is executed through one or more target functions, and each assessment subtask of the dike risk assessment task is sent to the second target resource node based on the third target strategy, including: determining the second target resource node based on the second task allocation strategy; and deploying one or more target functions of the target algorithm to the second target resource node.

[0013] Optionally, based on the dike hazard inspection task and the first and second state information, the target UAV and the first target resource node are controlled to perform the dike hazard inspection task, including: based on the dike hazard inspection task and the first and second state information, calling a first target strategy and a second target strategy from a preset strategy library, the first target strategy being used to guide the UAV to perform the dike hazard inspection task, and the second target strategy being used to guide the resource node to perform the dike hazard inspection task; controlling the target UAV to perform the dike hazard inspection task based on the first target strategy; determining the first target resource node based on the second target strategy, obtaining network connectivity information sent from multiple resource nodes, the network connectivity information being used to indicate the current network status of multiple resource nodes; determining whether to adjust the second target strategy based on the network connectivity information and the second state information to obtain the final second target strategy; determining the latest first target resource node based on the final second target strategy, and performing the dike hazard inspection task based on the latest first target resource node.

[0014] Optionally, the drone is equipped with one or more of the following: lidar, visible light camera, and thermal infrared imager; the dike patrol data includes one or more of the following: lidar point cloud data, visible light image, and thermal infrared data; and / or, the dike patrol results include one or more of the following: dike patrol data, patrol time, patrol location, flood control map of the patrol location, meteorological and hydrological data of the patrol location, drone flight route thumbnail, and drone flight mileage.

[0015] Optionally, the first target resource node has a preset data transmission order. After controlling the target UAV and the first target resource node to perform the dike hazard inspection task and obtain the dike inspection results based on the dike inspection data, the method further includes: controlling the last first target resource node to output the dike inspection results to the user server.

[0016] According to another aspect of the present invention, an automated inspection method for dike hazards based on vehicle-mounted unmanned aerial vehicles (UAVs) is also provided. This method is applied to the computing service layer of a fusion computing system, which is connected to the fusion adaptation layer described above. The method includes: acquiring user-inputted information and multi-source data collected for a target dike, the multi-source data including one or more of lidar point cloud data, visible light images, thermal infrared data, flood control maps, meteorological data, and hydrological data, the target dike being at least part of the dike inspected by the UAVs; preprocessing the input information and multi-source data to obtain preprocessed data, and extracting features from the preprocessed data to obtain feature data; using a preset intent library to perform intent perception and multi-attribute combination on the feature data to obtain feature processing results; matching the feature processing results with various preset intents in the intent library to obtain the target intent; and translating the target intent into business requirements and sending them to the fusion adaptation layer, the business requirements including dike hazard inspection tasks.

[0017] According to another aspect of the present invention, a user server is also provided. Including the computing service layer as described above, the user server further includes one or more of the following: a data overview module, a scheduling platform, a data center, and a system management module; the data overview module is used to receive and output dike patrol results from the self-fusion adaptation layer; the scheduling platform is used to receive input information from the user and send it to the computing service layer; the data center is used to record the historical mission execution records and historical dike patrol results of the UAV for user viewing, and the historical mission execution records include one or more of the following: the UAV that performed the historical dike hazard patrol mission, the execution time of the historical dike hazard patrol mission, and the execution mode of the historical dike hazard patrol mission; the system management module is used to set the user server's usage permissions and to record the real-time usage status and real-time location of the UAV.

[0018] Optionally, the user server may also include one or more of the following: a task receiving and allocation module, a field data acquisition module, a risk verification and assessment module, an information feedback and reporting module, a navigation and positioning service module, an emergency response and communication module, a user feedback and support module, and a data synchronization and cloud service module; the task receiving and allocation module is used to receive and display manual dike patrol tasks assigned by the system management module; the field data acquisition module is used to respond to the user's selection of a data acquisition tool, and to call the selected data acquisition tool to collect field dike data; the risk verification and assessment module is used to instruct the user to conduct dike risk verification according to a preset verification process, and also to provide a preset assessment template for the user to record dike risk point information, including dike risk type, dike risk level, and dike risk result. The system provides one or more types of information; the information feedback and reporting module is used to feed back dike feedback information to the system management module, including one or more of the following: text descriptions, images, videos, and standardized reports; the navigation and positioning service module is used to output the user's real-time location and manual dike patrol route to guide the user to the target dike patrol location; the emergency response and communication module is used to communicate in real time with the system management module and / or other devices using the user server; the user feedback and support module is used to receive information from users when using the user server, including user experience information and user suggestions; the data synchronization and cloud service module is used to store on-site dike data and dike feedback information in the cloud, and also to support cross-device access and updates of tasks assigned by the system management module, including manual dike patrol tasks.

[0019] The aforementioned technical solution, through coordinated scheduling of computing network resources, dynamically matches the real-time status of UAVs with the available computing network resources of distributed resource nodes and the resources required for dike patrol tasks. This significantly improves task response speed and resource utilization, helping to ensure the rapid mobilization of available UAVs and computing network resources during dike patrol missions. Furthermore, by selecting a primary target resource node with sufficient transmission capacity, low-latency, high-reliability transmission is achieved when the volume of dike patrol data is large, providing stable data support for subsequent hazard analysis. In addition, this solution enables a fully automated closed-loop process from task issuance and data collection to result feedback, effectively improving the accuracy and timeliness of dike patrols.

[0020] The above description is merely an overview of the technical solution of the present invention. In order to better understand the technical means of the present invention and to implement it in accordance with the contents of the specification, and in order to make the above and other objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention are described below. Attached Figure Description

[0021] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0022] Figure 1 A functional structure diagram of a fusion computing system according to an embodiment of the present invention is shown; Figure 2 A schematic flowchart of an automated dike hazard inspection method based on a vehicle-mounted unmanned aerial vehicle (UAV) according to an embodiment of the present invention is shown. Figure 3 A schematic diagram of the architecture of a fusion computing system according to an embodiment of the present invention is shown; Figure 4 A schematic flowchart illustrating the use of a fusion computing system to perform dike hazard inspection and assessment tasks according to an embodiment of the present invention is shown. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0024] To at least partially address the aforementioned technical problems, embodiments of the present invention provide an automated dike hazard inspection method and user server based on vehicle-mounted drones. This solution enables automated drone inspections based on vehicle-mounted drone nests, and utilizes the collaborative scheduling of multi-source heterogeneous sensing devices to achieve full coverage of dike inspections and multi-dimensional data collection, thereby improving the initiative, accuracy, timeliness, and intelligence of dike hazard response and facilitating the automation of dike inspections.

[0025] To facilitate the description and understanding of the technical solution of this invention, a converged computing system integrating computing and communication is first introduced. Please refer to... Figure 1The diagram illustrates the functional structure of a fusion computing system according to an embodiment of the present invention. The fusion computing system can be divided into a computing service layer, a fusion adaptation layer, and a computing network component layer. The computing service layer is typically deployed in a central control system, such as a city-level UAV intelligent dispatch platform or a bureau-level UAV cloud control platform. The computing service layer can abstract various tasks according to performance requirements. These abstracted tasks can be categorized into "services" as basic units, such as "piping identification service," "image transmission service," "levee hazard inspection service," and "levee hazard assessment service." Users can input business requests to select the corresponding service. Each service can correspond to a task to be executed and corresponding task indicators (e.g., service quality, experience quality). The computing service layer can translate each service from business language into machine-readable machine language, and the translated service can be distributed to the fusion adaptation layer. In this embodiment, the computing service layer can at least have a vehicle-mounted UAV platform module, an emergency communication coverage function module, and a flattened dispatch and visualization command function module. The vehicle-mounted UAV platform module can control the power supply (charging), take-off, landing, and recovery of the UAV, and can also transmit levee inspection results over the network. The emergency communication coverage module can formulate protection strategies and related algorithms (such as federated reinforcement learning algorithms) to ensure data security. It can also achieve multimodal communication through the integration of heterogeneous networks and automated networking through topology management and routing rules, grouping resource nodes into one or more clusters, each capable of executing corresponding preset tasks. The flattened scheduling and visual command module can receive user service requests in real time for remote command, receive data collected by sensing terminals (such as image acquisition devices on drones) for situational awareness, issue analysis tasks to analyze the collected data, and support receiving risk verification data uploaded from user servers. The automated dike hazard inspection method based on vehicle-mounted drones in this embodiment of the invention mainly implements the functions of the vehicle-mounted drone platform module and can optionally implement the functions of the flattened scheduling and visual command module. The computing service layer can communicate with the fusion adaptation layer through a northbound interface and can send service requests to the fusion adaptation layer.

[0026] For example, the fusion adaptation layer can be viewed as a serverless model (also known as a "serverless platform"). In Figure 1In the illustrated embodiment, the serverless model controller architecture can consist of an inter-domain controller and three intra-domain controllers. The inter-domain controller can communicate with each intra-domain controller via an east-west interface, and the intra-domain controllers can connect to the computing network component layer via a south-facing interface (P4Runtime). The computing network component layer can include one or more network domains, such as space-based networks, air-based networks, and ground-based networks. Each network domain can include multiple resource nodes, and each network domain can communicate with one intra-domain controller. The intra-domain controller can obtain the available computing network resources of each resource node within the communicated network domain, and the inter-domain controller can determine the available computing network resources of each resource node within each network domain through the intra-domain controller. Furthermore, the inter-domain controller can intelligently schedule resource nodes within each network domain for data transmission based on business needs and the available computing network resources of each resource node, and can also optionally perform data processing. Resource nodes can include, for example, cloud servers, edge servers, and terminal devices. This intelligent scheduling of resource nodes can achieve end-edge-cloud collaboration. In some embodiments, the resource nodes invoked can belong to different network domains, enabling cross-network domain resource invocation. Because the inter-domain controller takes into account the task metrics in the business requirements when scheduling resource nodes ( Figure 1 Displayed as QoS / QoE assurance, QoS stands for Quality of Service, and QoE stands for Quality of Experience. The fusion adaptation layer can complete UAV flight path planning, adaptive transmission scheduling for each resource node, and computation offloading, provided that the scheduled resource nodes meet the task objectives. "Computation offloading" refers to the fusion adaptation layer's ability to decompose computational tasks (primarily referring to dike hazard assessment tasks) into one or more virtual network functions when the tasks to be executed in the business requirements include computational tasks. These virtual network functions are then offloaded to one or more resource nodes for execution. It is understood that the resource nodes invoked include those capable of executing virtual network functions. In other words, the fusion adaptation layer can be viewed as using "network functions" as the basic unit, primarily responsible for scheduling computational network resources.

[0027] For example, the computing network component layer can be viewed as having "components" as the basic units, primarily responsible for data acquisition, storage, transmission, and computation. The computing network component layer may include vehicle-mounted drones as sensing and execution nodes. Figure 1(Not shown in the diagram) The number of vehicle-mounted drones can be one or more (usually multiple), and each vehicle-mounted drone can have a vehicle-mounted drone nest. The computing network component layer can also include multiple resource nodes, each of which can be a computing device / network device / integrated computing network device. The computing device can be composed of one or more processing units selected from CPU, ASIC, FPGA, GPU, NPU, TPU, etc. The network device can be a router / switch / gateway / server / P4 / controller, etc. The integrated computing network device can be a router or ground server configured with a computing power module, etc. In this embodiment of the invention, each vehicle-mounted drone can integrate a communication module for supporting data transmission (e.g., cellular network module, wireless LAN module, dedicated radio link module, satellite communication module, etc.), and the vehicle-mounted drone can optionally be equipped with a computing power module for supporting data computing (e.g., microcontroller, CPU, etc.). It can be understood that the communication module on the drone belongs to the network device, and the computing power module belongs to the computing device; therefore, the communication module and computing power module on the drone also belong to the resource nodes in the computing network component layer. The network device can have data transmission function and can also optionally have data storage function. Integrated computing and networking devices can have data computing and data transmission (i.e., forwarding) functions, and optionally data storage functions. Computing devices can have computing functions and optionally data storage functions. Each resource node in the computing and networking component layer can support multiple heterogeneous network protocols such as IPv4 and IPv6, thereby enabling data transmission between resource nodes. This resource node scheduling method can realize not only edge-cloud collaborative computing paradigms, but also intra-network and extra-network computing paradigms. The edge-cloud collaborative computing paradigm can be referred to in the aforementioned description of "edge-cloud collaboration," which will not be repeated here. The intra-network computing paradigm refers to lightweight computing subtasks that can be executed by the scheduled resource nodes. The extra-network computing paradigm refers to sending data transmitted by the scheduled resource nodes to data centers / supercomputing centers outside the network with high computing power to perform more complex computing tasks on the data. It is understandable that, when setting up a network, data centers / supercomputing centers with high computing power can also be used as resource nodes in the computing network component layer. When the resource nodes in the service chain include data centers / supercomputing centers, data centers / supercomputing centers can usually be used as the last resource node (also known as the terminal resource node) for data transmission.

[0028] In one specific embodiment, the converged computing system can integrate Software Defined Networking (SDN), P4 technology, and Network Function Virtualization (NFV) technology. SDN technology decouples the converged adaptation layer from the computing network component layer, allowing the converged adaptation layer to centrally perceive and schedule available computing network resources in the computing network component layer. Through the synergy of SDN and P4 technologies, virtual network functions that resource nodes in the computing network component layer can be flexibly configured to execute. Furthermore, the converged adaptation layer can implement in-band telemetry, load balancing, and rerouting. In addition, through SDN and NFV technologies, computing tasks can implement traditional network functions through virtual network functions, migrating from dedicated hardware to general-purpose devices. In one specific embodiment, the computing service layer and the converged service layer can interact using HTTP technology. The converged adaptation layer schedules resource nodes in the computing network component layer using RPC technology, and the computing network component layer can achieve data transmission between resource nodes using RDMA technology.

[0029] Please see Figure 2 The diagram shown is a schematic flowchart of an automated dike hazard inspection method based on a vehicle-mounted drone according to an embodiment of the present invention. According to one aspect of the present invention, an automated dike hazard inspection method based on a vehicle-mounted drone is provided, applied to the fusion adaptation layer of a fusion computing system, the method comprising steps S210-S230.

[0030] In step S210, the computing service layer of the self-fusion computing system obtains the dike hazard inspection task. The task content of the dike hazard inspection task includes controlling at least some drones to carry out dike inspection and receiving dike inspection data collected by at least some drones during the dike inspection process.

[0031] For example, the vehicle-mounted drone's nest has automatic charging, automatic take-off and landing, and automatic recovery functions, and it can move with the vehicle. The drone integrates a communication module (see above description) and a data acquisition device, such as a lidar, visible light camera, or thermal infrared imager. This data acquisition device can collect dike patrol data, which may include one or more of the following: raw visible light images, thermal infrared images, or laser point cloud images collected by the drone. The drone may also optionally carry a computing module (see above description). The dike hazard patrol service may include dike hazard patrol tasks, and optionally, task indicators corresponding to those tasks. The task content of the dike hazard patrol task includes controlling at least some drones in the computing network component layer, which is communicatively connected to the fusion adaptation layer in the fusion computing system, to conduct dike patrols. During the dike patrol process, the drone can transmit the collected dike patrol data to the fusion adaptation layer through resource nodes in the computing network component layer. The tasks of dike hazard inspection can also include receiving dike inspection data collected by drones during the dike inspection process.

[0032] In step S220, first status information and second status information are obtained. The first status information is used to indicate the usage status of the UAV, and the second status information is used to indicate the available computing network resources provided by each of the preset multiple resource nodes. Each resource node of at least some of the multiple resource nodes includes a network device or a computing network integrated device. Each resource node of the multiple resource nodes is used at least to transmit dike patrol data. The available computing network resources include available network resources used to support data transmission.

[0033] For example, the first state information may include, for instance, the flight time, current operating status, location, and sensor status information of each UAV in the computing network component layer. The second state information may indicate the available computing network resources provided by each resource node in the computing network component layer. For example, the explanation of "resource node" can be found in the foregoing embodiments and will not be repeated here. Each resource node can communicate with one or more other resource nodes, and multiple resource nodes can form a "computing network". The available computing network resources of a resource node include at least available network resources, which can support data transmission. Each resource node may have available network resources, which may be zero or non-zero. It should be noted that a resource node with zero available network resources cannot send data, but can receive data from other resource nodes it communicates with; a resource node with non-zero available network resources can send data to resource nodes it communicates with, and can also receive data from other resource nodes it communicates with. Specifically, for each resource node in the computing network, which includes at least some of the resource nodes, the resource node may include a network device or a computing network integrated device. Both the network device and the computing network integrated device have data transmission and reception functions. That is, for each resource node including a network device or a computing network integrated device, the available network resources of that resource node are non-zero. The second state information can be a quantization parameter used to quantify the available computing network resources. In some embodiments, the quantization parameter of the available network resources of each resource node may include the bandwidth of the associated link, end-to-end latency, latency jitter, packet loss rate, and throughput, etc. In other embodiments, when the resource node is a base station, the second state information can be a parameter used to quantify the base station signal quality.

[0034] In step S230, based on the dike hazard inspection task and the first and second status information, the target drone and the first target resource node are controlled to perform the dike hazard inspection task, so as to obtain the dike inspection results based on the dike inspection data. The target drone is a drone that is in use and can support the dike hazard inspection task, and the first target resource node is a resource node with corresponding available computing network resources that can support the transmission of dike inspection results.

[0035] For example, "The usage status can support the dike hazard inspection task" can mean that the usage status indicated by the first status information of the UAV can meet the requirements of the dike hazard inspection task. "The available network resources can support the transmission of dike inspection results" can mean that the amount of available network resources of the resource node meets the network resources required for the dike hazard inspection task. The first target resource node includes at least the communication module on the target UAV.

[0036] The aforementioned technical solution, through coordinated scheduling of computing network resources, dynamically matches the real-time status of UAVs with the available computing network resources of distributed resource nodes and the resources required for dike patrol tasks. This significantly improves task response speed and resource utilization, helping to ensure the rapid mobilization of available UAVs and computing network resources during dike patrol missions. Furthermore, by selecting a primary target resource node with sufficient transmission capacity, low-latency, high-reliability transmission is achieved when the volume of dike patrol data is large, providing stable data support for subsequent hazard analysis. In addition, this solution enables a fully automated closed-loop process from task issuance and data collection to result feedback, effectively improving the accuracy and timeliness of dike patrols.

[0037] Optionally, the dike hazard inspection task includes a drone inspection task and a node transmission task. Based on the dike hazard inspection task and first and second state information, the target drone and the first target resource node are controlled to perform the dike hazard inspection task to obtain dike inspection results based on the dike inspection data. This includes: based on the dike hazard inspection task and the first and second state information, calling a first target strategy and a second target strategy from a preset strategy library. The first target strategy includes a first task allocation strategy, a drone flight path planning strategy, and a drone task execution order strategy for the dike hazard inspection task. The second target strategy... The target strategy includes a transmission strategy for dike patrol results. The first task allocation strategy is used at least to allocate patrol positions for UAVs. Based on the first target strategy, each patrol sub-task of the UAV patrol task is sent to the nest of the target UAV that performs the patrol sub-task to control the target UAV to execute the corresponding patrol sub-task. Based on the second target strategy, the target flow table for node transmission tasks is sent to the first target resource node to control the first target resource node to form a transmission path that can transmit dike patrol results. The dike patrol data transmitted through the transmission path is obtained, wherein the dike patrol results include dike patrol data.

[0038] For example, the fusion adaptation layer can have a strategy library. The strategy library can pre-store various UAV scheduling strategies, which may include UAV route planning strategies, task allocation strategies, and execution order strategies. In a specific embodiment, the UAV route strategy may include a zigzag route for piping inspection, a route along the dike for dike crest crack inspection, and a circumferential scanning route for key areas. The task allocation strategy may include a block allocation strategy based on geographical regions, a preemptive allocation strategy based on task priority, etc. The execution order strategy may include strategies that execute according to the severity of the hazard, strategies that execute according to the shortest path principle, etc. Similarly, the strategy library can also pre-store various network transmission strategies. In a specific embodiment, the network transmission strategy may include a low-latency real-time video stream transmission strategy, a high-reliability high-definition photo return strategy, and an economical transmission strategy for periodic telemetry data, etc. After receiving a dike hazard inspection task, the fusion adaptation layer can combine the acquired first and second state information and match them in the strategy library. The fusion adaptation layer can then call the first target strategy most suitable for the UAV inspection task and the second target strategy most suitable for the node transmission task. In step S230, the fusion adaptation layer can allocate the UAV patrol task into multiple specific patrol sub-tasks according to the first task allocation strategy in the first target strategy, and assign them to the target UAVs in the best state. The first task allocation strategy can specify the patrol location of the UAVs to perform the dike hazard patrol task, and the patrol sub-tasks can indicate the task execution order and flight path of the corresponding target UAVs. The "patrol sub-tasks" are not sent directly to the UAVs, but to the nest to which each target UAV belongs. The nest, as the hosting base station for the UAVs, is responsible for the final task instruction conversion, takeoff preparation, and charging maintenance. For example, the fusion adaptation layer can be regarded as an SDN controller, which can generate one or more target flow tables according to the transmission strategy in the second target strategy. The flow table is a set of network rules that can specify the forwarding path, priority, and processing method of the dike patrol data. The fusion adaptation layer can distribute the target flow tables to the first target resource node (the first target resource node can be, for example, a network device or computing network integrated device that supports protocols such as OpenFlow). After receiving the flow table, the first target resource node can forward the dike patrol results according to its rules, thus forming a dynamic transmission path for transmitting the dike patrol results. The target drone performs patrols according to the planned route, transmitting the collected dike patrol data to the fusion adaptation layer through the established transmission path, and optionally to the data processing center / cloud platform. Exemplarily, but not limitingly, the data processing center / cloud platform can analyze and process the received dike patrol results (such as AI-powered intelligent identification of hazards, manual assessment), ultimately generating data processing results containing information such as the location, type, and level of the hazard.

[0039] The above technical solution, by invoking the first target strategy, can automatically decompose and allocate drone patrol tasks. This is beneficial for improving the response speed and execution efficiency of dike patrols. By invoking the second target strategy and sending the target flow table to the selected first target resource node, it helps to dynamically construct one or more transmission paths for transmitting dike patrol data, which can effectively avoid data congestion, loss, or delay, and help users receive dike patrol results in real time and completely.

[0040] Optionally, the dike hazard inspection task is represented by computing power identification information and computing power behavior description information. The computing power identification information indicates the dike hazard inspection task, and the computing power behavior description information indicates the computing network resources required for the task. The first state information includes the UAV's flight time information. The second state information includes population identification information, population behavior description information, node identification information, and node behavior description information. The population identification information indicates the corresponding resource node population, and the population behavior description information indicates the preset tasks that the corresponding resource node population can perform. These preset tasks include node transmission tasks. The node identification information indicates the corresponding resource node, and the node behavior description information... This is used to indicate the transmission subtasks that the corresponding resource nodes can execute; based on the dike hazard inspection task and the first and second state information, the first and second target strategies are invoked from the preset strategy library, including: invoking the first target strategy from the strategy library based on the endurance information, computing power identification information, and computing power behavior description information; determining the target group that can execute the dike hazard inspection task based on the computing power identification information, computing power behavior description information, group identification information, and group behavior description information, the target group includes at least two resource nodes; and invoking the second target strategy from the strategy library based on the computing power behavior description information, the node identification information and node behavior description information corresponding to the target group.

[0041] Please see Figure 3The diagram illustrates the architecture of a converged computing system according to an embodiment of the present invention. The converged computing system can be divided into an entity domain, a sensing and control domain, and a knowledge domain. In the entity domain, the computing network component layer can identify each resource node using component identification information (i.e., node identity, NID). The convergence adaptation layer can pre-divide resource nodes with functional or attribute associations into groups, and identify each group using group identity information (Family Identity, FID). Resource nodes within a group can collaborate to execute corresponding preset tasks. The computing service layer can identify each service using computing power identification information (Computing Identity, CID), and correspondingly, identify tasks within each service. Collaboration between groups can realize corresponding services. In the sensing and control domain, the computing network component layer defines the behavioral characteristics (including executable transmission subtasks, transmission performance, etc.) of a single resource node using component behavior description information (i.e., node behavior description information, NBD). The fusion and adaptation layer defines the functions of corresponding groups through Family Behavior Description (FBD), i.e., the preset tasks that the group can perform. The computing service layer defines the computing network resources required for corresponding services through Computing Service Behavior Description (CBD), and can also define other resources required for corresponding services (such as task duration, task distance, and required device power). The behavioral characteristics of resource nodes can be clustered to form the functions (i.e., behaviors) of groups. Matching the functions of groups with the computing network resources required for services can achieve the matching of computing power demand and supply. In the knowledge domain, the network component model library of the computing network component layer stores the state information of each resource node, and the group management strategy library of the fusion and adaptation layer stores network transmission strategies and computing task allocation strategies for defining strategies and rules for group collaboration. It can also store drone scheduling strategies to guide the linkage and scheduling of drone groups (each drone can be divided into multiple drone groups, and each drone group can include one or more drones). The computing power management knowledge base of the computing service layer stores the task types of each service and the computing network resources required by them.

[0042] For example, a dike hazard inspection task can be defined using computing power identification information (indicating the task type) and computing power behavior description information (indicating the required resources). In the first state information, the drone's endurance information can indicate the endurance of each drone. In the second state information, the population identification information and population behavior description information can jointly define the corresponding resource node population and the preset tasks that the corresponding resource node population can perform. Node identification information and node behavior description information can jointly define the corresponding resource node and the transmission sub-tasks that the corresponding resource node can perform. For example, by matching the drone's endurance with the drone inspection task and the required resources, a first target policy can be invoked in the policy library. By traversing the population identification information and population behavior description information of all resource node populations, the preset tasks that each resource node population can perform can be determined. Matching the computing power identification information and computing power behavior description information with the traversed information can be regarded as a "task-resource" matching process, which can determine the target population that can perform the dike hazard inspection task. By traversing the node identification information and node behavior description information of each resource node in the target population, the transmission subtasks that each resource node in the target population can execute can be determined. Based on the computing power behavior description information and the traversed information, the second target strategy can be called in the strategy library.

[0043] In one specific embodiment, the computing power identification information indicates a dike hazard inspection task, and the computing power behavior description information indicates the need for real-time transmission of high-definition images. The task takes 1 hour, covers a 3-kilometer dike section, and requires a bandwidth of 40Mbps per resource node. The fusion adaptation layer obtains the flight duration information of each drone. Based on the flight duration information, computing power identification information, and computing power behavior description information, the first target strategy invoked includes drones with a flight duration greater than or equal to 1.2 hours (slightly longer than the task duration) that can perform the dike hazard inspection task. The fusion adaptation layer can also obtain the group identification information and group behavior description information of each resource node group. If the group identification information of a certain resource node group is an image transmission group, and the group behavior description information indicates support for multi-resource node collaboration, a bandwidth of more than 50Mbps for a single resource node, and a node deployment range that can cover a 3-kilometer dike section, then this resource node group can be identified as the target group. The fusion adaptation layer can also obtain the node identification information and node behavior description information of each resource node in the target group, thereby invoking a second target strategy from the strategy library. The second target strategy could be to instruct resource nodes 1, 2, and 3 in the target population to execute node transmission tasks sequentially.

[0044] The above technical solution, through hierarchical matching of "computing power behavior description - population behavior description - node behavior description", can accurately align the computing power requirements of dike patrol with the capabilities of computing network resources, which helps to ensure the execution of dike hazard patrol tasks.

[0045] Optionally, the patrol subtask can indicate the target flight path, target attitude, and sensor data collected by the corresponding target UAV.

[0046] For example, the target flight path can instruct the target UAV to take off from point A and fly to point B along a preset spatial path (e.g., a zigzag path, a path along the dike, etc.). The target attitude can indicate the pitch angle, roll angle, and heading angle of the target UAV. Sensors can include, for example, lidar, visible light cameras, and thermal infrared imagers. Taking a visible light camera as an example, its acquisition parameters can include acquisition mode (timed / distance shooting), resolution, focal length, etc. LiDAR acquisition parameters can include, for example, point cloud density and scanning frequency. Thermal infrared imager acquisition parameters can include, for example, temperature range and emissivity. Through the patrol sub-task, executable actions and acquisition methods can be specified for the target UAV, thereby ensuring that the target UAV can collect reliable dike patrol data.

[0047] Optionally, the fusion computing system further includes a computing network component layer, which includes multiple resource nodes. The computing network component layer is deployed with a first neural network for processing dike patrol data to obtain dike hazard assessment results. The first neural network includes a second neural network. The method further includes: acquiring the mission execution log of the target UAV and adding the dike patrol data and mission execution log to a preset database to update the target training samples in the database. The computing network component layer is used to train the second neural network indicated by the mission execution log based on the dike patrol data in the target training samples; and / or, acquiring the latest available computing network resources of at least one of the multiple resource nodes and updating the acquired latest available computing network resources to a preset database. The database is used to store the available computing network resources of multiple resource nodes. The fusion adaptation layer acquires the available computing network resources of multiple resource nodes from the database.

[0048] For example, when the target UAV performs a patrol sub-task, it can transmit the original dike patrol data through the transmission path formed by the first target resource node, and it can also transmit the task execution log. The task execution log can indicate the second neural network in the computing network component layer used to process the dike patrol data. The second neural network is a lightweight network in the first neural network used to process the dike patrol data. In some embodiments, the computing power module on the target UAV can deploy the second neural network, and the task execution log can indicate the second neural network deployed by the computing power module on the UAV. The fusion adaptation layer can have a preset database, and the received task execution log and dike patrol data can be added to the database to update the target training samples in the database. The resource node on the computing network component layer with the second neural network deployed can train the second neural network based on the dike patrol data in the target training samples under the control of the fusion adaptation layer. Specifically, when the resource node with the second neural network deployed does not perform the relevant task, the fusion adaptation layer can use the dike patrol data as input based on the corresponding available computing resources and use manual annotation to iteratively train the second neural network. For example, at least some resource nodes in the computing network component layer can periodically (e.g., every 30 seconds) report their own status data to the fusion adaptation layer. This status data can indicate the latest available computing network resources for the corresponding resource node. Alternatively, the fusion adaptation layer can periodically send query commands to each resource node to obtain the latest available computing network resources for each resource node. Upon receiving the latest available computing network resources, the historical status data of the corresponding resource node in the database can be automatically overwritten, thereby ensuring that the available computing network resources stored in the database are real-time data. In step S120, the second status information is the information obtained by the fusion adaptation layer from the database.

[0049] The above technical solution associates dike patrol data with task execution logs to form target training samples and updates them to the database. As the sample size accumulates and training is iterated, the performance of the second neural network can be gradually improved. By dynamically updating the available computing resources of each resource node, it can be ensured that the available computing resources of the resource nodes obtained by the fusion adaptation layer are the latest data, which helps to avoid scheduling to resource nodes that are overloaded or faulty, and reduces problems such as transmission delay and task interruption caused by resource mismatch.

[0050] Optionally, the method further includes: obtaining a dike hazard assessment task from the computing service layer, wherein the task content of the dike hazard assessment task includes processing dike patrol data to obtain a dike hazard assessment result, wherein each resource node of at least some of the multiple resource nodes includes a computing device or a computing network integrated device, and the available computing network resources also include available computing resources for supporting data processing; and controlling a second target resource node to execute the dike hazard assessment task according to the dike hazard assessment task and the second state information to obtain the corresponding dike hazard assessment result, wherein the second target resource node is a resource node whose corresponding available computing network resources can support the processing of dike patrol data.

[0051] For example, the tasks obtained from the computing service layer may also include dike hazard assessment tasks, the content of which includes processing dike patrol data. The available computing network resources of a resource node may also include available computing resources, which can support data computation. Similarly, each resource node may have available computing resources, which can be zero or non-zero. It is understood that a resource node with zero available computing resources cannot participate in executing the dike hazard assessment task; a resource node with non-zero available computing resources can participate in executing the dike hazard assessment task. Specifically, for each resource node in the fused computing system (at least some of the resource nodes), the resource node may include a computing device or a computing network integrated device, both of which have data computation capabilities. That is, for each resource node including a computing device or a computing network integrated device, the available computing resources of that resource node are non-zero. The quantifiable parameters of available computing resources for each resource node can include, for example, floating-point operations per second (FLOPS), storage read / write speed (IOPS), Boolean operations per second (BOPS), and logical operations per second (LOPS). "Available computing resources can support the processing of dike patrol data" indicates that the amount of available computing resources for a resource node meets the computing resources required for the dike emergency response task. It should be noted that the second target resource node can belong to at least some of the first target resource nodes, or it can be the same as only some of the resource nodes in the first target resource node.

[0052] The above technical solution, by dynamically scheduling resource nodes to perform the task of assessing the danger of dikes, helps to avoid concentrating all the massive data computing tasks on a small number of computing devices / integrated computing and network devices, and can make full use of the available computing resources of resource nodes. This helps to achieve the global optimal allocation of the task of assessing the danger of dikes based on the available computing resources of each resource node, and can significantly improve the execution efficiency of the task of assessing the danger of dikes.

[0053] Optionally, based on the dike hazard assessment task and the second state information, the second target resource node is controlled to execute the dike hazard assessment task, including: based on the dike hazard assessment task and the second state information, calling a third target strategy from a preset strategy library, the third target strategy including the target algorithm required for the dike hazard assessment task and the second task allocation strategy, the target algorithm being the algorithm used to process dike patrol data; and sending each assessment sub-task of the dike hazard assessment task to the second target resource node based on the third target strategy, so that the second target resource node can execute the corresponding assessment sub-task respectively.

[0054] For example, the strategy library can pre-store various algorithms for hazard assessment (such as classification algorithms, segmentation algorithms, target detection algorithms, etc.) and computational task offloading strategies (such as proximity processing strategies, high-load task priority allocation to the cloud strategies, and lightweight task priority allocation to edge node strategies, etc.). After receiving the levee hazard assessment task, the fusion adaptation layer can match the acquired second state information with the algorithms and computational task offloading strategies in the strategy library. The fusion adaptation layer can then call the third target strategy that best suits the levee hazard assessment task. The third target strategy can include the target algorithm required for the levee hazard assessment task and the second task allocation strategy, which defines the rules for allocating the levee hazard assessment task to different resource nodes. Based on the third target strategy, the levee hazard assessment task can be decomposed into one or more assessment sub-tasks and sent to the second target resource nodes for execution of the corresponding assessment sub-tasks.

[0055] The above technical solution can decompose the large computing power required for a dike hazard assessment task into multiple assessment sub-tasks, each requiring a smaller computing power. This allows a dike hazard assessment task to be allocated to multiple resource nodes. This task allocation method can adapt to the currently available computing resources in the computing network component layer, has high rationality, and can significantly improve the resource utilization rate of each resource node.

[0056] Optionally, the target algorithm is executed through one or more target functions, and each assessment subtask of the dike risk assessment task is sent to the second target resource node based on the third target strategy, including: determining the second target resource node based on the second task allocation strategy; and deploying one or more target functions of the target algorithm to the second target resource node.

[0057] For example, the target algorithm can consist of one or more objective functions, and the subtasks can be executed through the corresponding objective functions. After determining the second target resource nodes based on the second task allocation strategy, the objective functions of the target algorithm can be deployed to each of the second target resource nodes. For example, but not limitingly, the second state information can also indicate the functional functions that have been deployed / cached on each resource node. If the determined second target resource nodes already have corresponding objective functions deployed / cached, the fusion adaptation layer does not need to be redeployed. The above technical solution, by deploying the objective functions of the target algorithm to each of the second target resource nodes, helps to realize distributed computing for dike hazard tasks, thereby helping to achieve edge-cloud collaboration by combining edge-cloud computing power.

[0058] Optionally, based on the dike hazard inspection task and the first and second state information, the target drone and the first target resource node are controlled to perform the dike hazard inspection task, including: based on the dike hazard inspection task and the first and second state information, calling a first target strategy and a second target strategy from a preset strategy library, the first target strategy being used to guide the drone to perform the dike hazard inspection task, and the second target strategy being used to guide the resource node to perform the dike hazard inspection task; controlling the target drone to perform the dike hazard inspection task based on the first target strategy; determining the first target resource node based on the second target strategy, and obtaining network connectivity information sent from multiple resource nodes, the network connectivity information being used to indicate the current network status of multiple resource nodes; determining whether to adjust the second target strategy based on the network connectivity information and the second state information to obtain the final second target strategy; determining the latest first target resource node based on the final second target strategy, and performing the dike hazard inspection task based on the latest first target resource node.

[0059] For example, the specific explanations of the first and second target strategies can be found in the foregoing embodiments, and will not be repeated here. The fusion adaptation layer can control the target UAV to perform the dike hazard inspection task based on the first target strategy, and determine the first target resource node based on the second target strategy. The first target resource node determined this time can form an initial transmission path. During the process of the target UAV performing the dike hazard inspection task, the fusion adaptation layer can obtain the network connectivity information sent by each resource node. The network connectivity information can indicate the current network status of each resource node (e.g., node connectivity, inter-node latency, link bandwidth utilization, packet loss rate, link status, etc.). If the network connectivity information indicates that the current network status meets the transmission requirements, the second target strategy does not need to be adjusted. If the network connectivity information indicates that the current network status does not meet the transmission requirements, such as a sudden increase in latency of the initial transmission path / insufficient available bandwidth / resource node failure / non-optimal transmission path, the second target strategy can be adjusted to obtain the final second target strategy. Using the final second target strategy, the first target resource node can be re-determined, and the fusion adaptation layer can send the new target pointer to the latest first target resource node to construct a new transmission path. The results of the dike patrol can continue to be transmitted along the newly constructed transmission path.

[0060] The above technical solution can ensure the smooth transmission of dike patrol results by changing the second target strategy and re-determining the first target resource node when the determined transmission path cannot meet the transmission requirements.

[0061] Optionally, the drone is equipped with one or more of the following: lidar, visible light camera, and thermal infrared imager; the dike patrol data includes one or more of the following: lidar point cloud data, visible light image, and thermal infrared data; and / or, the dike patrol results include one or more of the following: dike patrol data, patrol time, patrol location, flood control map of the patrol location, meteorological and hydrological data of the patrol location, drone flight route thumbnail, and drone flight mileage.

[0062] For example, the metadata corresponding to the dike patrol data can include patrol time and location. The fusion adaptation layer can also obtain drone operational data from the drone's flight control system, including drone flight path thumbnails and drone flight mileage. Flood control maps, meteorological data, and hydrological data for the patrol locations can be obtained from the geographic information system via external API interfaces. The obtained dike patrol data is rich in data type and has strong reliability and reference value.

[0063] Optionally, the first target resource node has a preset data transmission order. After controlling the target UAV and the first target resource node to perform the dike hazard inspection task and obtain the dike inspection results based on the dike inspection data, the method further includes: controlling the last first target resource node to output the dike inspection results to the user server.

[0064] For example, after step S130, the last first target resource node can also send the dike patrol results to the user server for the user to view. This ensures that the user can obtain the dike patrol results in a timely manner through the user server.

[0065] According to another aspect of the present invention, an automated inspection method for dike hazards based on vehicle-mounted unmanned aerial vehicles (UAVs) is also provided. This method is applied to the computing service layer of a fusion computing system, which is connected to the fusion adaptation layer described above. The method includes: acquiring user-inputted information and multi-source data collected for a target dike, the multi-source data including one or more of lidar point cloud data, visible light images, thermal infrared data, flood control maps, meteorological data, and hydrological data, the target dike being at least part of the dike inspected by the UAVs; preprocessing the input information and multi-source data to obtain preprocessed data, and extracting features from the preprocessed data to obtain feature data; using a preset intent library to perform intent perception and multi-attribute combination on the feature data to obtain feature processing results; matching the feature processing results with various preset intents in the intent library to obtain the target intent; and translating the target intent into business requirements and sending them to the fusion adaptation layer, the business requirements including dike hazard inspection tasks.

[0066] For example, the description of the computing service layer can be referred to in the foregoing embodiments, and will not be repeated here. The description of multi-source data can be referred to in the foregoing embodiments regarding dike patrol data, and will not be repeated here. User input information can be a business request in natural language form, such as "dike patrol and hazard inspection". The computing service layer can acquire multi-source data and user input information, and use a preprocessor to preprocess the multi-source data and input information (e.g., denoising, registration, fusion, etc.). Feature extraction can be performed on the preprocessed data to obtain feature data. The computing service layer has a preset intent library, which can predefine various intents related to dike patrol (such as routine dike section patrol, key inspection of hazard points, hazard assessment, etc.). Intent recognition technologies such as rule matching and machine learning can be used to perceive the intent in the feature data. The perceived intent is filled and integrated according to the attributes of each dimension corresponding to the preset attribute template to form a standardized structure, thus obtaining the feature processing result. Mapping and matching the feature processing result with various preset intents in the intent library yields the target intent. Intent translation technology can convert target intent into standardized information, which may include, for example, the computing power behavior description information and computing power identification information described in the aforementioned embodiments.

[0067] In the above technical solution, the computing service layer can accurately map business needs to computing power needs based on real-time perceived multi-source data and user input information through intent translation. This can decouple the business layer (computing service layer) from the implementation layer (algorithm component layer), making the system easier to expand and maintain.

[0068] Please see Figure 4 The diagram shown is a schematic flowchart illustrating the use of a fusion computing system to perform dike hazard inspection and assessment tasks according to an embodiment of the present invention. Figure 4 In the illustrated embodiments, the computing service layer may include traffic collection components (e.g., NetFLOW, DPDK), a preprocessor, and an intent library. The fusion adaptation layer may include a policy library, a database, an orchestrator, and a controller. The computing network component layer may include a vehicle-mounted drone, an edge server, and a border gateway. A lightweight drone platform (which can be considered as the flight control system of the aforementioned embodiments) and sensors integrated on the drone can access the fusion computing system according to preset access rules (including IP, port, protocol, etc.), and multi-source data can be collected using the drone and its integrated sensors. Figure 4 The illustrated embodiment refers to multi-source data as multi-source high-throughput dam data. The traffic collection component can receive user input (i.e., intent requests in natural language). Furthermore, the multi-source data collected in real-time by the UAV can be transmitted in real-time to the traffic collection component of the computing service layer according to the transmission path defined by the controller of the fusion adaptation layer (this process is...). Figure 4(Not shown in the image). The traffic collection component can send the received input information and multi-source data to the preprocessor in real time. The preprocessor performs preprocessing and feature extraction based on the real-time collected user intent and multi-source data to obtain feature data, and packages the feature data and sends it to the intent library. The intent library can perform fuzzy intent analysis on the feature data, that is, perform intent perception and multi-attribute combination on the feature data to obtain feature processing results. The intent library can obtain the target intent by matching the feature processing results with a variety of stored preset intents, translate the target intent into service requirements, and send it to the policy library of the fusion adaptation layer. Service requirements (i.e., the business requirements of the aforementioned embodiments) are described by computing power identification information (CID) and computing power behavior description information (CBD). Specifically, business requirements can be described by computing power identification information and computing power behavior description information. Computing power identification information can indicate the dike hazard inspection task, computing power behavior description information can indicate the computing network resources required by the task, and computing power behavior description information can be predefined by the QoS requirements and / or QoE requirements of the dike hazard inspection task. QoS requirements and / or QoE requirements are typically associated with the drone's endurance, levee terrain, pedestrian traffic in the levee area, interference sources, high points, risk priority, etc.

[0069] Please continue reading. Figure 4As shown, after receiving a service request, the policy library can retrieve the currently stored first and second state information from the database. The first state information includes the drone's remaining flight time (referring to the current remaining flight time), and the second state information includes group identifier (FID), group behavior description (FBD), node identifier (NID), and node behavior description (NBD). The second state information can, for example, indicate base station signal quality, edge server CPU / memory usage, and deployed / cached subtasks and functions on the edge server. Based on the received service request and the first and second state information, the policy library can call the corresponding policy interface to generate an inspection task allocation policy (i.e., the first task allocation policy in the aforementioned embodiment), a drone flight path planning policy, a drone task execution order policy, and a routing policy (i.e., the transmission policy in the aforementioned embodiment). The policy library can send inspection task allocation policies, UAV flight path planning policies, and UAV task execution order policies to the orchestrator. The orchestrator can then use the API server to issue inspection sub-tasks to the respective nests of each target UAV based on these policies. The nests then control the UAVs to execute the inspection sub-tasks, which can specify the planned flight path, attitude, and shooting parameters of the corresponding target UAV. The policy library can also translate routing policies into control flow tables (i.e., the aforementioned target flow tables) and issue them to the border gateway (i.e., the first target resource node in the aforementioned embodiment), thereby forming an end-to-end programmable network control path (i.e., the transmission path in the aforementioned embodiment). The border gateway can transmit multi-source data collected by the UAVs to the controller and can also transmit the UAVs' task execution logs to the controller. The controller can add the task execution logs and associated dike inspection data to the database to update the target training samples in the database. Exemplarily but not limitingly, the border gateway can also transmit network connectivity information to the controller, which can dynamically adjust the transmission policy based on the network connectivity information. In addition, the controller can optionally obtain the latest available computing network resources of each resource node and update the database with the latest available computing network resources.

[0070] Please continue reading. Figure 4 As shown, when the service requirements also include the task of assessing the risk of flooding in dikes, the orchestrator can call the third-target strategy in the strategy library via API. The third-target strategy includes the target algorithm used for assessment, and the target algorithm consists of one or more strategy functions. The third-target strategy also includes a function allocation strategy (i.e., the aforementioned second task allocation strategy). The orchestrator can deploy each strategy function to each second-target resource node based on the function allocation strategy. The second-target resource node can belong to any computing device / computing-network integrated device in the edge-cloud. Figure 4In the illustrated embodiment, the second target resource node may include an edge server and a computing module on the drone. The second target resource node can intelligently analyze and provide risk warnings for the dike patrol data, obtaining a dike hazard assessment result for the target dike, and transmitting it to the controller. The controller can send the dike hazard assessment result to the user. The dike hazard assessment result includes one or more of the following: patrol time / location, data size, hazard type, flight path thumbnail, flight mileage, latitude and longitude coordinates of suspected piping points, image, and image name.

[0071] According to another aspect of the present invention, a user server is also provided. Including the computing service layer as described above, the user server further includes one or more of the following: a data overview module, a scheduling platform, a data center, and a system management module; the data overview module is used to receive and output dike patrol results from the self-fusion adaptation layer; the scheduling platform is used to receive input information from the user and send it to the computing service layer; the data center is used to record the historical mission execution records and historical dike patrol results of the UAV for user viewing, and the historical mission execution records include one or more of the following: the UAV that performed the historical dike hazard patrol mission, the execution time of the historical dike hazard patrol mission, and the execution mode of the historical dike hazard patrol mission; the system management module is used to set the user server's usage permissions and to record the real-time usage status and real-time location of the UAV.

[0072] For example, the data overview module provides users with a centralized and intuitive interface capable of real-time monitoring and viewing of the operational status and key data of the entire computing network component layer. The interface displayed in the data overview module can include the logos of various management organizations, real-time weather information, time information, and 2D or 3D simulation maps. The 2D or 3D simulation maps can display the distribution of drones and their nests, as well as the geographical distribution of drones currently executing tasks. Furthermore, the data overview module can display flight records and logistical statistics for all drones, providing detailed statistical information including the number of online devices, flight counts, number of tasks in progress, number of completed tasks, total number of anomalies, number of incomplete tasks, number of failed tasks, total flight mileage, and total flight time. The data overview module also features first-person view monitoring of drones, displaying real-time dike patrol data collected by drones, as well as image processing results (belonging to dike hazard assessment results) after processing the dike patrol data. Processing algorithms include, for example, target detection and classification algorithms. Through these comprehensive displays and statistics, the data overview module enables managers to quickly grasp the overall system operation, make timely decisions, and ensure the smooth progress of inspection tasks and proper maintenance of equipment.

[0073] For example, the scheduling platform interacts with users, specifically receiving user input information to schedule drones. Furthermore, the platform includes a device list menu and a flight path planning menu. In the device list menu, users can view detailed distribution and status information of their owned drone nests and drones. In the flight path planning menu, users can edit various types of flight paths online, including regional routes, preset routes, and routes customized to specific needs. The scheduling platform also supports setting up electronic fences, allowing users to draw restricted areas to ensure drone flight safety. In addition, the platform can display mission routes on 2D / 3D simulated maps and simulate real-time drone flight trajectories, providing real-time drone information, including key parameters such as longitude, latitude, flight status, speed, altitude, and battery level. Through the scheduling platform, users can manually or automatically control drones to perform tasks, create, edit, query, and manage flight paths, facilitating fully autonomous inspection and intelligent three-dimensional collaborative control of aircraft. This optimizes the allocation of inspection resources and improves inspection efficiency and response speed.

[0074] For example, the data center possesses functions such as task execution status recording, data storage, analysis, and playback. The recorded information includes key details such as the equipment used to perform the task, the execution time, and the execution mode (e.g., comprehensive patrol, group-based patrol, relay patrol, rotation patrol, manual foot patrol, human-machine collaborative patrol, fixed-point automated monitoring, integrated air-ground patrol, etc.). In drone-based dike patrol and inspection missions, the data center not only stores multimedia data such as images and videos collected by the drones but also supports online playback and download of this data. Users can retrieve and analyze historical inspection data through the data center. Furthermore, the data center has a data retrieval function, allowing users to query based on conditions such as task execution time, task status, and execution route name to quickly locate the required data. The data center can provide reliable data management and analysis tools for drone-based dike patrol and inspection missions, helping managers better understand the dike status, assess risks, formulate response measures, and optimize inspection strategies, thereby improving the efficiency and accuracy of inspection work.

[0075] For example, the system management module includes an organization management submodule, a role management submodule, an equipment management submodule, and a user management submodule. The organization management submodule is used by administrators on the user server to create and maintain organizational units based on the company or department structure, enabling unified management of inspection teams at the organizational level. For each organization, the organization management submodule allows setting access permissions to protect data security and prevent unauthorized access. The role management submodule allows administrators to configure access permissions corresponding to job responsibilities. The equipment management submodule manages drones, recording key drone information (such as model and manufacturer) and synchronizing drone usage status and location in real-time on a 2D / 3D simulation map. The equipment submodule also handles adding, modifying, and removing equipment. The user management submodule maintains all accounts within the organization, including adding new users, deleting users, modifying user information and passwords, and querying users.

[0076] Optionally, the user server may also include one or more of the following: a task receiving and allocation module, a field data acquisition module, a risk verification and assessment module, an information feedback and reporting module, a navigation and positioning service module, an emergency response and communication module, a user feedback and support module, and a data synchronization and cloud service module; the task receiving and allocation module is used to receive and display manual dike patrol tasks assigned by the system management module; the field data acquisition module is used to respond to the user's selection of a data acquisition tool, and to call the selected data acquisition tool to collect field dike data; the risk verification and assessment module is used to instruct the user to conduct dike risk verification according to a preset verification process, and also to provide a preset assessment template for the user to record dike risk point information, including dike risk type, dike risk level, and dike risk result. The system provides one or more types of information; the information feedback and reporting module is used to feed back dike feedback information to the system management module, including one or more of the following: text descriptions, images, videos, and standardized reports; the navigation and positioning service module is used to output the user's real-time location and manual dike patrol route to guide the user to the target dike patrol location; the emergency response and communication module is used to communicate in real time with the system management module and / or other devices using the user server; the user feedback and support module is used to receive information from users when using the user server, including user experience information and user suggestions; the data synchronization and cloud service module is used to store on-site dike data and dike feedback information in the cloud, and also to support cross-device access and updates of tasks assigned by the system management module, including manual dike patrol tasks.

[0077] For example, in the system management module, administrators within an organization can assign tasks to users, including manual dike patrols. The user assigned the task can be determined based on the user's location and the priority of the task. Users can view currently pending tasks through the task list provided by the task receiving and assignment module, as well as task details and urgency levels. The field data acquisition module has built-in various data acquisition tools, including GPS positioning tools, image acquisition tools, video acquisition tools, and audio recording tools. Users can use these tools to collect relevant field data, such as images / videos / audio reflecting topography, water level changes, and potential risk points. The risk verification and assessment module provides a standardized verification process and assessment templates. The templates require information such as dike risk type, dike risk level, and dike risk outcome. Users can edit and / or upload dike feedback information through the information feedback and reporting module, and can also upload dike feedback information to the system management module. The information feedback and reporting module also provides standardized verification report templates. The navigation and positioning service module integrates navigation and positioning technologies, enabling it to output the user's real-time location and manual patrol route via map navigation to guide the user to the target patrol location. Furthermore, the navigation and positioning service module can receive navigation guidance information from the system management module. In the event of a sudden emergency or in need of support, the user can send an emergency distress signal to the system management module and / or the user server via the emergency response and communication module, and can also communicate in real-time with the system management module and / or other devices using the user server.

[0078] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of the invention. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the invention as claimed in the appended claims.

[0079] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0080] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed.

[0081] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0082] Similarly, it should be understood that, in order to streamline the invention and aid in understanding one or more of the various aspects of the invention, features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the description of exemplary embodiments of the invention. However, this approach should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the corresponding claims, its inventive point lies in solving the corresponding technical problem with fewer features than all of those in a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0083] Those skilled in the art will understand that, apart from the mutual exclusion of features, all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or elements of any method or apparatus so disclosed may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0084] Furthermore, those skilled in the art will understand that although some embodiments herein include certain features included in other embodiments but not others, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the claims, any of the claimed embodiments can be used in any combination.

[0085] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. The present invention can also be implemented as an apparatus program (e.g., a computer program and computer program product) for performing some or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium, or can take the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0086] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0087] The above are merely specific embodiments or descriptions of the present invention, and the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for automated inspection of dike hazards based on vehicle-mounted unmanned aerial vehicles (UAVs), characterized in that, The method, applied to the fusion adaptation layer of a fusion computing system, includes: The task of inspecting dike hazards is obtained from the computing service layer of the fusion computing system. The task of inspecting dike hazards includes controlling at least some drones to conduct dike inspections and receiving dike inspection data collected by at least some drones during the dike inspection process. First status information and second status information are obtained. The first status information is used to indicate the usage status of the UAV. The second status information is used to indicate the available computing network resources provided by each of the preset multiple resource nodes. Each resource node of at least some of the multiple resource nodes includes a network device or a computing network integrated device. Each resource node of the multiple resource nodes is at least used to transmit dike patrol data. The available computing network resources include available network resources for supporting data transmission and available computing power resources for supporting data processing. Based on the dike hazard inspection task, the first status information, and the second status information, the target drone and the first target resource node are controlled to perform the dike hazard inspection task, so as to obtain the dike inspection result based on the dike inspection data. The target drone is a drone whose usage status can support the dike hazard inspection task, and the first target resource node is a resource node whose corresponding available computing network resources can support the transmission of the dike inspection result.

2. The method according to claim 1, characterized in that, The dike hazard inspection task includes a drone inspection task and a node transmission task. The step of controlling the target drone and the first target resource node to execute the dike hazard inspection task based on the dike hazard inspection task, the first status information, and the second status information, to obtain dike inspection results based on the dike inspection data, includes: Based on the dike hazard inspection task and the first and second status information, the first target strategy and the second target strategy are invoked from the preset strategy library. The first target strategy includes the first task allocation strategy, the UAV flight path planning strategy, and the UAV task execution order strategy for the dike hazard inspection task. The second target strategy includes the transmission strategy for the dike inspection results. The first task allocation strategy is at least used to allocate the inspection position of the UAV. Based on the first target strategy, each patrol sub-task of the UAV patrol mission is sent to the nest of the target UAV that performs the patrol sub-task, so as to control the target UAV to perform the corresponding patrol sub-task. Based on the second target strategy, the target flow table for the node transmission task is sent to the first target resource node to control the first target resource node to form a transmission path that can transmit the dike inspection results; Acquire the dike patrol data transmitted through the transmission path, wherein the dike patrol results include the dike patrol data.

3. The method according to claim 2, characterized in that, The dike hazard inspection task is represented by computing power identification information and computing power behavior description information. The computing power identification information is used to indicate the dike hazard inspection task, and the computing power behavior description information is used to indicate the computing network resources required for the dike hazard inspection task. The first status information includes the UAV's endurance information. The second status information includes population identification information, population behavior description information, node identification information, and node behavior description information. The population identification information is used to indicate the corresponding resource node population, and the population behavior description information is used to indicate the preset tasks that the corresponding resource node population can perform. The preset tasks include the node transmission tasks. The node identification information is used to indicate the corresponding resource node, and the node behavior description information is used to indicate the transmission sub-tasks that the corresponding resource node can perform. The step of invoking a first target strategy and a second target strategy from a preset strategy library based on the dike hazard inspection task and the first and second status information includes: Based on the battery life information, the computing power identification information, and the computing power behavior description information, the first target policy is invoked from the policy library; The target group capable of performing the dike hazard inspection task is determined based on the computing power identification information, the computing power behavior description information, the group identification information, and the group behavior description information. The target group includes at least two resource nodes. The second target strategy is invoked from the strategy library based on the computing power behavior description information, the node identification information corresponding to the target group, and the node behavior description information.

4. The method according to claim 2, characterized in that, The patrol subtask can indicate the target flight path, target attitude, and the data collected by the sensors mounted on the target UAV.

5. The method according to claim 1, characterized in that, The fusion computing system further includes a computing network component layer, which includes the plurality of resource nodes. The computing network component layer is deployed with a first neural network for processing the dike patrol data to obtain dike risk assessment results. The first neural network includes a second neural network. The method further includes: The task execution log of the target UAV is obtained and the dike patrol data and the task execution log are added to a preset database to update the target training samples in the database. The computing network component layer is used to train the second neural network indicated by the task execution log based on the dike patrol data in the target training samples. And / or, The latest available computing network resources of at least one of the multiple resource nodes are obtained, and the obtained latest available computing network resources are updated to a preset database. The database is used to store the available computing network resources of the multiple resource nodes. The fusion adaptation layer obtains the available computing network resources of the multiple resource nodes from the database.

6. The method according to any one of claims 1-5, characterized in that, The method further includes: The task of assessing the danger of dikes is obtained from the computing service layer. The task of assessing the danger of dikes includes processing the dike patrol data to obtain the assessment result of the danger of dikes. Each resource node of at least some of the multiple resource nodes includes a computing device or a computing network integrated device. The available computing network resources also include available computing resources for supporting data processing. Based on the dike hazard assessment task and the second status information, the second target resource node is controlled to execute the dike hazard assessment task in order to obtain the corresponding dike hazard assessment result. The second target resource node is a resource node whose corresponding available computing network resources can support the processing of the dike patrol data.

7. The method according to claim 6, characterized in that, The step of controlling the second target resource node to execute the dike hazard assessment task based on the dike hazard assessment task and the second status information includes: Based on the dike hazard assessment task and the second status information, a third target strategy is invoked from the preset strategy library. The third target strategy includes the target algorithm required for the dike hazard assessment task and the second task allocation strategy. The target algorithm is the algorithm used to process the dike patrol data. Based on the third objective strategy, each assessment sub-task of the dike risk assessment task is sent to the second objective resource node, so that the second objective resource node can execute the corresponding assessment sub-task respectively.

8. The method according to claim 7, characterized in that, The target algorithm is executed through one or more target functions. The step of sending each assessment sub-task of the dike risk assessment task to the second target resource node based on the third target strategy includes: The second target resource node is determined based on the second task allocation strategy; Deploy one or more objective functions of the target algorithm to the second target resource node.

9. The method according to any one of claims 1-5, characterized in that, The step of controlling the target drone and the first target resource node to perform the dike hazard inspection task based on the dike hazard inspection task, the first status information, and the second status information includes: Based on the dike hazard inspection task and the first and second status information, the first and second target strategies are invoked from the preset strategy library. The first target strategy is used to guide the UAV to perform the dike hazard inspection task, and the second target strategy is used to guide the resource node to perform the dike hazard inspection task. Based on the first target strategy, the target UAV is controlled to perform the dike hazard inspection task; The first target resource node is determined based on the second target strategy, and network connectivity information sent from the multiple resource nodes is obtained. The network connectivity information is used to indicate the current network status of the multiple resource nodes. Based on the network connectivity information and the second status information, determine whether to adjust the second target strategy to obtain the final second target strategy; Based on the final second target strategy, the latest first target resource node is determined, and the dike hazard inspection task is executed based on the latest first target resource.

10. The method according to any one of claims 1-5, characterized in that, The drone is equipped with one or more of the following: lidar, visible light camera, and thermal infrared imager; the dike patrol data includes one or more of the following: lidar point cloud data, visible light images, and thermal infrared data; and / or, The dike patrol results include one or more of the following: dike patrol data, patrol time, patrol location, flood control map of the patrol location, meteorological and hydrological data of the patrol location, drone flight path thumbnail, and drone flight mileage.

11. The method according to any one of claims 1-5, characterized in that, The first target resource node has a preset data transmission order. After the control target UAV and the first target resource node perform the dike hazard inspection task to obtain the dike inspection result based on the dike inspection data, the method further includes: Control the last first target resource node to output the dike patrol results to the user server.

12. A method for automated inspection of dike hazards based on vehicle-mounted unmanned aerial vehicles (UAVs), characterized in that, A computing service layer applied to a fusion computing system, the computing service layer being connected to a fusion adaptation layer as described in any one of claims 1-11, the method comprising: The system acquires user input information and multi-source data collected for the target levee. The multi-source data includes one or more of the following: lidar point cloud data, visible light images, thermal infrared data, flood control maps, meteorological data, and hydrological data. The target levee is the levee that at least part of the drone patrols. The input information and the multi-source data are preprocessed to obtain preprocessed data, and the preprocessed data is then subjected to feature extraction to obtain feature data. The feature data is processed using a pre-defined intent library to perform intent perception and multi-attribute combination to obtain feature processing results. The feature processing result is matched with various preset intents in the intent library to obtain the target intent; The target intent is translated into business requirements and sent to the fusion adaptation layer, and the business requirements include the dike hazard inspection task.

13. A user server, characterized in that, Including the computing service layer as described in claim 12, the user server further includes: One or more of the following modules: data overview module, scheduling platform, data center, and system management module; The data overview module is used to receive and output the dike inspection results from the fusion adaptation layer. The scheduling platform is used to receive input information from the user and send it to the computing service layer; The data center is used to record the historical mission execution records and historical dike patrol results of the UAV for users to view. The historical mission execution records include one or more of the following: the UAV that performed the historical dike hazard patrol mission, the execution time of the historical dike hazard patrol mission, and the execution mode of the historical dike hazard patrol mission. The system management module is used to set the usage permissions of the user server and to record the real-time usage status and real-time location of the drone.

14. The user server according to claim 13, characterized in that, The user server also includes one or more of the following: a task receiving and allocation module, a field data acquisition module, a risk verification and assessment module, an information feedback and reporting module, a navigation and positioning service module, an emergency response and communication module, a user feedback and support module, and a data synchronization and cloud service module. The task receiving and allocation module is used to receive manual dike patrol tasks allocated by the system management module and output and display them; The on-site data acquisition module is used to respond to the user's selection operation of the data acquisition tool, and call the data acquisition tool selected by the selection operation to collect on-site dike data; The risk verification and assessment module is used to instruct users to conduct dike risk verification according to a preset verification process, and is also used to provide a preset assessment template for users to record dike risk point information. The dike risk point information includes one or more of the following: dike risk type, dike risk level, and dike risk result. The information feedback and reporting module is used to feed back dike feedback information to the system management module. The dike feedback information includes one or more of the following: text description information, pictures, videos, and standardized reports. The navigation and positioning service module is used to output the user's real-time location and manual dike patrol route to guide the user to the target dike patrol location. The emergency response and communication module is used to communicate in real time with the system management module and / or other devices using the user server. The user feedback and support module is used to receive information from users when using the user server. The feedback information includes user experience information and user usage suggestions. The data synchronization and cloud service module is used to store the on-site dike data and the dike feedback information in the cloud, and is also used to support cross-device access to and updating of tasks assigned by the system management module, including manual dike patrol tasks.

Images